Reputation: 225
I'm new to Elixir and Phoenix framework. How can I dynamically add parameter to ecto query ? For example:
def index(conn, _params) do
departments = Repo.all(Department)
render(conn, "index.json", data: departments)
end
Is it possible to use the _params in Ecto query ? Something like:
Repo.get_by(Department, _params)
Upvotes: 1
Views: 1392
Reputation: 428
I know this thread is old, but one can also make use of Ecto.Changeset.cast
, and also benefit of its validation. For example, to select all:
with %Ecto.Changeset{valid?: true} = changeset <- Ecto.Changeset.cast(%Department{}, params, [:department_name]) do
clauses = Enum.to_list(changeset.changes)
query =
from e in Department,
where: ^clauses,
order_by: [desc: e.inserted_at]
{:ok, Repo.all(query)}
else
changeset -> {:error, changeset}
end
Upvotes: 0
Reputation: 9079
you can get stuff dynamically ofcourse, Repo.get_by in the second argument expects Keyword.t | Map.t
, either a keyword or a map. You can set it directly from params into a Map or keyword, and pass that into the query.
Example
def index(conn, _params) do
query_conditions = [id: params["id"], name: params["department_name"] ]
departments = Repo.get_by(Department, query_conditions)
render(conn, "index.json", data: departments)
end
As @Dogbert stated in the comments that get_by
will get you 0 or 1 result, more results will raise an error. I only used it because you used get_by
in the question.
Upvotes: 1
Reputation: 222040
You can use "bindingless queries" for this, but there's a problem. Ecto expects the column names to be atoms and the argument to be a keyword list. The latter is easy, but converting arbitrary user input into atoms may result in the Erlang VM crashing. Here's a safe way to do this using a whitelist of column names:
# user input
params = %{"first_name" => "John", "last_name" => "Smith", "age" => 28}
filtered_params =
params
|> Map.take(~w(first_name last_name age))
|> Enum.map(fn {k, v} -> {String.to_atom(k), v} end)
from(MyApp.Person, where: ^filtered_params)
|> MyApp.Repo.all
|> IO.inspect
Output:
[debug] SELECT p0."id", p0."first_name", p0."last_name", p0."age", p0."inserted_at", p0."updated_at" FROM "people" AS p0 WHERE (((p0."age" = ?) AND (p0."first_name" = ?)) AND (p0."last_name" = ?)) [28, "John", "Smith"] OK query=6.5ms queue=14.8ms
[%MyApp.Person{__meta__: #Ecto.Schema.Metadata<:loaded>, age: 28,
first_name: "John", id: 1, inserted_at: #Ecto.DateTime<2016-06-11T16:01:49Z>,
last_name: "Smith", updated_at: #Ecto.DateTime<2016-06-11T16:01:49Z>}]
Upvotes: 4